Chapter

Logic for Programming, Artificial Intelligence, and Reasoning

Volume 3835 of the series Lecture Notes in Computer Science pp 322-336

Deciding Separation Logic Formulae by SAT and Incremental Negative Cycle Elimination

  • Chao WangAffiliated withCarnegie Mellon UniversityNEC Laboratories America
  • , Franjo IvančićAffiliated withCarnegie Mellon UniversityNEC Laboratories America
  • , Malay GanaiAffiliated withCarnegie Mellon UniversityNEC Laboratories America
  • , Aarti GuptaAffiliated withCarnegie Mellon UniversityNEC Laboratories America

* Final gross prices may vary according to local VAT.

Get Access

Abstract

Separation logic is a subset of the quantifier-free first order logic. It has been successfully used in the automated verification of systems that have large (or unbounded) integer-valued state variables, such as pipelined processor designs and timed systems. In this paper, we present a fast decision procedure for separation logic, which combines Boolean satisfiability (SAT) with a graph based incremental negative cycle elimination algorithm. Our solver abstracts a separation logic formula into a Boolean formula by replacing each predicate with a Boolean variable. Transitivity constraints over predicates are detected from the constraint graph and added on a need-to basis. Our solver handles Boolean and theory conflicts uniformly at the Boolean level. The graph based algorithm supports not only incremental theory propagation, but also constant time theory backtracking without using a cumbersome history stack. Experimental results on a large set of benchmarks show that our new decision procedure is scalable, and outperforms existing techniques for this logic.